Sustainable sugarcane bagasse-derived activated carbon for comparative Cr(VI) and Co(II) removal: experimental and ANN modeling integration.

Developing sustainable and high-performance adsorbents for heavy metal removal remains a major environmental challenge. In this study, sugarcane bagasse was valorized into activated carbon (AC-SCB) through a low-temperature sulfuric acid activation process, providing an energy-efficient and environmentally friendly adsorbent for wastewater treatment. The prepared material exhibited a mesoporous structure with a specific surface area of 251.7 ± 12 m2/g and abundant oxygen-containing surface functionalities, promoting strong interactions with metal ions. The novelty of this work lies in the comparative investigation of Cr(VI) and Co(II) adsorption under identical experimental conditions, combined with statistical and machine-learning modeling approaches. Batch adsorption experiments demonstrated excellent removal efficiencies of 99.14 ± 0.12% for Cr(VI) at pH 3.02 and 99.40 ± 0.10% for Co(II) at pH 6.21. The adsorption mechanisms were mainly governed by electrostatic interactions for Cr(VI) and surface complexation/chelation for Co(II). Kinetic and equilibrium studies revealed that the adsorption process followed the pseudo-second-order kinetic and Langmuir isotherm models, yielding maximum adsorption capacities of 895.2 ± 5.8 mg/g for Cr(VI) and 934.3 ± 6.1 mg/g for Co(II). Process optimization was performed using Response Surface Methodology (RSM), which provided highly accurate predictive models (RCr(VI) = 0.9953 ± 0.0004, MSECr(VI) = 0.0094 ± 0.0003; RCo(II) = 0.9996 ± 0.0002, MSECo(II) = 0.0008 ± 0.0001). Artificial neural network (ANN) and support vector machine (SVM) models were subsequently developed and compared to further improve prediction performance. Among the developed models, ANN achieved the highest predictive accuracy (RCr(VI) = 0.99625 ± 0.0005, MSECr(VI) = 0.0014 ± 0.0002; RCo(II) = 0.99986 ± 0.0001, MSECo(II) = 0.0011 ± 0.0001), outperforming both RSM and SVM (RCr(VI) = 0.992 ± 0.001, MSECr(VI) = 0.0045 ± 0.0003; RCo(II) = 0.997 ± 0.001, MSECo(II) = 0.0025 ± 0.0002) in capturing the nonlinear behavior of the adsorption system. Furthermore, AC-SCB demonstrated good regeneration performance, retaining 75 ± 2% of its initial adsorption efficiency after five adsorption-desorption cycles. The combined experimental, statistical, and machine-learning approaches demonstrate that low-temperature activated sugarcane bagasse is a sustainable, efficient, and scalable adsorbent for the treatment of heavy-metal-contaminated wastewater, while demonstrating the potential of integrating artificial intelligence tools with experimental approaches for advanced adsorption process prediction and optimization.